Novel approaches to inform tropical bird conservation in human modified landscapes

In this thesis I utilise a combination of newly advanced methodological and statistical approaches to assess knowledge gaps concerning biodiversity in human-modified tropical landscapes. Specifically, I use cutting-edge LiDAR technology, occupancy modelling and soundscape analysis to document the responses of tropical birds to land-use change in Borneo.

I first evaluate the contribution that riparian reserves - protected natural vegetation around waterways in production landscapes - have in supporting biodiversity. By assessing the avian community structure and richness of riparian reserves I demonstrate that these landscape features can offer significant biodiversity benefits, and support comparable levels of species diversity to logged riparian forests provided they are of sufficient size (>80 m in total width) and habitat quality (>75 tC ha-1 of tree biomass). I show that in oil palm estates riparian reserves would need to be >200 m in total width (i.e. 100 m from each riverbank) to preserve comparable numbers of forest specialist bird to logged riparian forest.

I then examine whether responses of species and trait groups to habitat disruption follow linear trajectories or non-linear responses whereby abrupt changes to occupancy and diversity occur once thresholds of disturbance are exceeded. Habitat disruption across a land-cover gradient from intact forest to oil palm plantations was characterised via LiDAR metrics that quantify habitat structure in three dimensions. By scrutinising the individual responses of 171 bird species and 17 different multi-species trait groups to these metrics via hierarchical multi-species occupancy modelling, I show that the majority of species respond to habitat degradation in a non-linear fashion. I demonstrate that thresholds in species response scale up to abrupt changes in trait group richness, particularly those associated with important ecosystem functions such as pollination, seed dispersal and insectivory. I find trait groups exhibit highly varied thresholds from one another. I also highlight how exceeding particular thresholds of degradation in human modified tropical landscapes could result in abrupt changes to ecosystem functioning, thereby making human-modified tropical landscape less resilient to further perturbations.

Last, I seek to test the application of recently developed acoustic approaches for monitoring biodiversity in human-modified tropical landscapes. I assess the performance of five commonly used 'soundscape' indices in corresponding to variation in observed or estimated bird diversity from field data. I find that sources of acoustic bias in production landscapes (including human produced noise and the sound of running), make broad application of acoustic monitoring technologies to heavily disturbed habitats such as intensive farmland challenging. I demonstrate that controlling for time-of-day, using noise-reduction algorithms and excluding certain habitat types, improves the capacity of acoustic indices to reflect both observed bird richness, and estimates of species numbers derived from occupancy models.

Taken together, the three studies in this thesis reveal the biodiversity value of riparian areas, the potential for non-linear responses of species to habitat change, and the efficacy of novel monitoring techniques applied to biodiversity monitoring in human-modified tropical landscapes. I offer a number of recommendations and applications of these three sets of findings and explore their implication for biodiversity conservation in tropical regions. By addressing these three knowledge gaps using a combination of newly available innovations I demonstrate not only the importance of the findings themselves, but also highlight how innovations in technology, analytical technique and monitoring approach when used in conjunction can elucidate biodiversity patterns that were otherwise less well known.